Page Comparison

...

Info
title Note
The GRASS engine used in AccessMod will resample raster format layers presenting a different resolution than the DEM to make it match the DEM’s resolution. The resampling technique used by the GRASS engine is the similar to a "nearest neighbor" one . This technique assigns (i.e.. assigning the value in each "new" pixel based on the value stored in the nearest "previous" pixel). However, GRASS is not using a dedicated "nearest neighbor" function in that case, but rather applies a "pixel mask" of the resolution of the DEM over any other raster, and assign the values of the other raster to match the resolution of the DEM. While using such technique might be appropriate when applied to the landcover raster format layer We were not able to specify exactly how GRASS proceeds to do that, so we should not assume the process is a clean "nearest neighbor" attribution. It is therefore strongly encouraged to always ensure that the input raster files have the exact same resolution. While change of raster resolution may likely be correct for categorical raster such as the the landcover, this is not the case for the population distribution layer as this transformation will not conserve the original total population and can therefore translate in wrong output statistics.

It is important to bear in mind that input vector data sets such as the road network and barriers to movement (e.g., rivers, bodies of water) are converted to grids when generating the merged land cover distribution grid. This conversion is done using the same resolution as the original land cover distribution grid. Depending on this resolution, the conversion can have a direct impact on the spatial relationship between the roads and the barriers in the merged land cover layer.

As a first example,

Caption ref
anchor artificial bridges

 below the figure below shows the result of the conversion (rasterization) of a road (black) and river (blue) layer for different resolutions. As we can see, the lower the resolution (i.e. larger raster grid size), the higher the risk to generate an overlap between roads and the rivers, therefore creating artificial “bridges” (red arrows) that do not exist in reality.

...

anchor	artificial
caption	Example of rasterization of a road (in black) and river (in blue) segments (a) at a resolution of 180 meters (b) and 1 kilometer (c) with the indication of artificial passages (red arrows) that gets created

...

1. A situation in which the density of artificial bridges to be corrected is low (as indicated in red in

   Caption ref
   anchor bridges

    belowthe figure below, road in green and river in white). In this case, the user should open the merged land cover, the road and the hydrographic network layers in a GIS software and:
   1. Generate a buffer with a radius equivalent to 1.75 time the resolution of the original land cover grid (in light blue in

      Caption ref
      anchor bridges

      b figure b below);
   2. Put the hydrographic network layer in the editing mode and move the concerned segments outside of the buffer area as presented in

      Caption ref
      anchor bridges

      c figure c below (highlighted blue line). You might also want to adjust the way the hydrographic network cuts the road network by placing them perpendicularly as also shown in

      Caption ref
      anchor bridges

      c figure c below;
   3. Save your edits, import the new layer in AccessMod and run the merge land cover tool once again. The new resulting merged land cover should then look like presented in

      Caption ref
      anchor bridges

      d figure d below.

   Image Removed

   CaptioneditemanchorbridgescaptionExample of correction process operated on the hydrographic network with low density of artificial bridges with rivers showing up in white and roads showing up in green (see text above for explanations)
   1.  

   Image Added

   2. A situation in which the density of vertex on the segments to be corrected is dense (indicated in red in

   Caption ref
   anchor hydr_correc

   a figure a below, road in green and river in white). In this case, open the merged land cover, the road and the hydrographic network layers in a GIS software and:
   1. Generate a buffer with a radius equivalent to 1.75 time the resolution of the original land cover grid (in blue in

      Caption ref
      anchor hydr_correc

      b figure b below)
   2. Put the hydrographic network layer in the editing mode and add a new river segment outside of the buffer area as presented in

      Caption ref
      anchor hydr_correc

      c figure c below (light blue line)
   3. Save your edits and run the first module of AccessMod. The new resulting merged land cover should then look like presented in

      Caption ref
      anchor hydr_correc

      d figure d below

...

anchor	hydr_correc
caption	Example of correction process operated on the hydrographic network with high density of vertex (see text above for explanations)

It is important to be aware that the choice of a coarse resolution can generate the merging of separate road segments that are close to each other. This is shown in

Caption ref
anchor raster_roads

below the figure below where there are artificial intersections between road segments at low resolution (larger grid size). Unfortunately, no correction can be made in this case.

...

...

Users also need to be conscious of the impact that the change in resolution has on the travel time in cell where patients are reaching the road network.

...

In addition to the above, the rasterization of elongated water body polygons (and other barriers to movement stored as polygons) does not always generate continuous surfaces, unlike when rasterizing vector lines. Figure 5 below provides an illustration of such a problem. As we can see in

Caption ref
anchor raster_elong

cfigure c below, working with lower resolution (larger grid sizes) will create discontinuities, therefore generating artificial bridges in these areas.

...

The solution to address this problem is to:

1. Convert your polygon water body layer into polyline.
2. Merge the polyline resulting from step 1 with your line format barrier layer (see section 3.3.1.8 for details).
3. Use both the original polygon layer and the merged line layer as barrier when creating the merged land cover layer in AccessMod.

At the same time, working with low resolution raster format data might lead to some health facilities to be located in a “No Data” area (river network for example). If this happens, the user will have to manually move the facility to the nearest cell. The user should be very careful to move the facility to the correct side of the river/water body. This modification of the location can be considered as an adjustment to the resolution used in the analysis, and its impact on the results of the modelling should typically be very small.

Accuracy

It is important to have a clear understanding of the level of accuracy of the different geospatial data used for the analysis. As can be seen in the figure below, a facility located on the wrong side of a river will lead to the design of a completely different catchment area and therefore contribute to a different population coverage.

Image Added

In view of the above, working at the highest resolution possible would seem to be the best option. Nonetheless, it is important to remember that increasing the resolution of the raster layers results in an increase of the size of the files and the required computational time (see section 3.2.5).

Accuracy is also a function of scale (i.e. the ratio of a distance on the map to the corresponding distance on the ground), which is itself directly linked to the resolution of the raster format data used in the analysis (Ebener, 2016). It is therefore important to ensure consistency from an accuracy perspective among the different layers being used in the analysis.